EX 3 Trabalho final word2vec.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\eder\\Anaconda3\\lib\\site-packages\\gensim\\utils.py:1197: UserWarning: detected Windows; aliasing chunkize to chunkize_serial\n",
      "  warnings.warn(\"detected Windows; aliasing chunkize to chunkize_serial\")\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import pickle\n",
    "import tensorflow as tf\n",
    "import gensim\n",
    "from gensim.models import Word2Vec\n",
    "import re\n",
    "import nltk\n",
    "from nltk.tokenize import word_tokenize\n",
    "from nltk.corpus import stopwords\n",
    "import random\n",
    "import string\n",
    "from sklearn.utils import shuffle\n",
    "import multiprocessing\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn import svm\n",
    "from sklearn import metrics\n",
    "from nltk.stem import WordNetLemmatizer\n",
    "from nltk.stem.rslp import RSLPStemmer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>fold_id</th>\n",
       "      <th>cv_tag</th>\n",
       "      <th>html_id</th>\n",
       "      <th>sent_id</th>\n",
       "      <th>text</th>\n",
       "      <th>tag</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>cv000</td>\n",
       "      <td>29590</td>\n",
       "      <td>0</td>\n",
       "      <td>films adapted from comic books have had plenty...</td>\n",
       "      <td>pos</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>cv000</td>\n",
       "      <td>29590</td>\n",
       "      <td>1</td>\n",
       "      <td>for starters , it was created by alan moore ( ...</td>\n",
       "      <td>pos</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>cv000</td>\n",
       "      <td>29590</td>\n",
       "      <td>2</td>\n",
       "      <td>to say moore and campbell thoroughly researche...</td>\n",
       "      <td>pos</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>cv000</td>\n",
       "      <td>29590</td>\n",
       "      <td>3</td>\n",
       "      <td>the book ( or \" graphic novel , \" if you will ...</td>\n",
       "      <td>pos</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>cv000</td>\n",
       "      <td>29590</td>\n",
       "      <td>4</td>\n",
       "      <td>in other words , don't dismiss this film becau...</td>\n",
       "      <td>pos</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   fold_id cv_tag  html_id  sent_id  \\\n",
       "0        0  cv000    29590        0   \n",
       "1        0  cv000    29590        1   \n",
       "2        0  cv000    29590        2   \n",
       "3        0  cv000    29590        3   \n",
       "4        0  cv000    29590        4   \n",
       "\n",
       "                                                text  tag  \n",
       "0  films adapted from comic books have had plenty...  pos  \n",
       "1  for starters , it was created by alan moore ( ...  pos  \n",
       "2  to say moore and campbell thoroughly researche...  pos  \n",
       "3  the book ( or \" graphic novel , \" if you will ...  pos  \n",
       "4  in other words , don't dismiss this film becau...  pos  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dfBase = pd.read_csv('datasets/movie_review.csv')\n",
    "dfBase.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>fold_id</th>\n",
       "      <th>cv_tag</th>\n",
       "      <th>html_id</th>\n",
       "      <th>sent_id</th>\n",
       "      <th>text</th>\n",
       "      <th>tag</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5</td>\n",
       "      <td>cv503</td>\n",
       "      <td>11196</td>\n",
       "      <td>30</td>\n",
       "      <td>perhaps sci-fi spectaculars are just not an ac...</td>\n",
       "      <td>neg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4</td>\n",
       "      <td>cv428</td>\n",
       "      <td>12202</td>\n",
       "      <td>66</td>\n",
       "      <td>how does a talking decapitated head show that ...</td>\n",
       "      <td>neg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>cv201</td>\n",
       "      <td>7421</td>\n",
       "      <td>7</td>\n",
       "      <td>she's the lone survivor this time .</td>\n",
       "      <td>neg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>cv199</td>\n",
       "      <td>9629</td>\n",
       "      <td>42</td>\n",
       "      <td>it just doesn't have the emotional impact that...</td>\n",
       "      <td>pos</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>7</td>\n",
       "      <td>cv768</td>\n",
       "      <td>11751</td>\n",
       "      <td>13</td>\n",
       "      <td>the one real problem i had with it is the unre...</td>\n",
       "      <td>pos</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   fold_id cv_tag  html_id  sent_id  \\\n",
       "0        5  cv503    11196       30   \n",
       "1        4  cv428    12202       66   \n",
       "2        2  cv201     7421        7   \n",
       "3        1  cv199     9629       42   \n",
       "4        7  cv768    11751       13   \n",
       "\n",
       "                                                text  tag  \n",
       "0  perhaps sci-fi spectaculars are just not an ac...  neg  \n",
       "1  how does a talking decapitated head show that ...  neg  \n",
       "2                she's the lone survivor this time .  neg  \n",
       "3  it just doesn't have the emotional impact that...  pos  \n",
       "4  the one real problem i had with it is the unre...  pos  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = shuffle(dfBase)\n",
    "df = df.reset_index(drop = True)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "stemmer = RSLPStemmer()\n",
    "\n",
    "def lemmatize_stemming(text):\n",
    "    return stemmer.stem(WordNetLemmatizer().lemmatize(text, pos='v'))\n",
    "\n",
    "def TreatText(data):\n",
    "    #import pdb; pdb.set_trace()\n",
    "    stops = set(stopwords.words(\"english\"))  # melhora a performance convertendo num set\n",
    "    data['text'] = [re.sub(\"[^a-zA-Z]\", \" \",data['text'][i]) for i in range(len(data))] #mantém apenas letras (há números, links, etc.)     \n",
    "    data['text'] = [word_tokenize(data['text'][i].lower()) for i in range(len(data))] # caixa baixa\n",
    "    data['text'] = [[lemmatize_stemming(w) for w in data['text'][i] if w not in stops]for i in range(len(data))]# lematize/stemming/remove stopwords\n",
    "    return(data['text'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0            [perhap, sci, fi, spectacul, act, medium]\n",
       "1    [talk, decapitat, head, show, te, dispondent, ...\n",
       "2                                   [lon, surviv, tim]\n",
       "3          [emot, impact, suspect, mr, cameron, shoot]\n",
       "4    [one, real, probl, unrealistic, natur, main, c...\n",
       "Name: text, dtype: object"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dfText = TreatText(df)\n",
    "dfText.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 0, ..., 1, 1, 1], dtype=int64)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labels = np.array(df['tag'].apply(lambda x: 1 if x == 'pos' else 0)) \n",
    "labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "dim_vec = 300\n",
    "min_count = 10\n",
    "window = 4\n",
    "num_workers = multiprocessing.cpu_count()\n",
    "seed = tf.set_random_seed(1713)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def meanVector(model,phrase):\n",
    "    vocab = model.wv.vocab\n",
    "    phrase = \" \".join(phrase)\n",
    "    phrase = [x for x in word_tokenize(phrase) if x in vocab]\n",
    "    #Quando não houver palavra o vector recebe 0 para todas as posições\n",
    "    if phrase == []:\n",
    "        vetor = [0.0]*dim_vec \n",
    "    else: \n",
    "        #Caso contrário, calculando a matriz da frase\n",
    "        vetor = np.mean([model[word] for word in phrase],axis=0)\n",
    "    return vetor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def createFeatures(base, modelo): \n",
    "    features = [meanVector(modelo,base[i])for i in range(len(base))]\n",
    "    return features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def generateF1Score(dfText, modelo):\n",
    "    dfeatures = createFeatures(dfText, modelo)\n",
    "    X_train, X_test, y_train, y_test = train_test_split(dfeatures[0:len(dfeatures)], labels[0:len(labels)], test_size=0.3,random_state=109)\n",
    "    clf = svm.SVC(kernel='linear') \n",
    "    clf.fit(X_train, y_train)\n",
    "    y_pred = clf.predict(X_test)\n",
    "    f_score = metrics.f1_score(y_test, y_pred, average='micro')\n",
    "    print(\"Accuracy:\",f_score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tamanho do vocabulário do Word2Vec CBOW:  7283\n"
     ]
    }
   ],
   "source": [
    "modeloCBOW = Word2Vec(dfText,\n",
    "                    min_count = min_count, \n",
    "                    size = dim_vec, \n",
    "                    window = window,\n",
    "                    seed = seed,\n",
    "                    workers = num_workers,\n",
    "                    sg = 0) #sg = 0 -> CBOW e sg = 1 -> skipgram\n",
    "print(\"Tamanho do vocabulário do Word2Vec CBOW: \", len(modeloCBOW.wv.vocab.keys()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\eder\\Anaconda3\\lib\\site-packages\\ipykernel_launcher.py:10: DeprecationWarning: Call to deprecated `__getitem__` (Method will be removed in 4.0.0, use self.wv.__getitem__() instead).\n",
      "  # Remove the CWD from sys.path while we load stuff.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.5627317676143387\n"
     ]
    }
   ],
   "source": [
    "generateF1Score(dfText, modeloCBOW)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tamanho do vocabulário do Word2Vec skipgram:  7283\n"
     ]
    }
   ],
   "source": [
    "modeloCSKIPGRAM = Word2Vec(dfText,\n",
    "                    min_count = min_count, \n",
    "                    size = dim_vec, \n",
    "                    window = window,\n",
    "                    seed = seed,\n",
    "                    workers = num_workers,\n",
    "                    sg = 1) #sg = 0 -> CBOW e sg = 1 -> skipgram\n",
    "print(\"Tamanho do vocabulário do Word2Vec skipgram: \", len(modeloCSKIPGRAM.wv.vocab.keys()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\eder\\Anaconda3\\lib\\site-packages\\ipykernel_launcher.py:10: DeprecationWarning: Call to deprecated `__getitem__` (Method will be removed in 4.0.0, use self.wv.__getitem__() instead).\n",
      "  # Remove the CWD from sys.path while we load stuff.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.5968788627935723\n"
     ]
    }
   ],
   "source": [
    "generateF1Score(dfText, modeloCSKIPGRAM)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## O Modelo skipgram teve uma performance superior ao CBOW( 0.5627317676143387 Vs 0.5968788627935723)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}